A sample-to-answer, portable platform for rapid detection of pathogens with a smartphone interface

Yu-Dong Ma a, Kuang-Hsien Li b, Yi-Hong Chen a, Yung-Mao Lee b, Shang-Ta Chou a, Yue-Yuan Lai a, Po-Chiun Huang b, Hsi-Pin Ma *b and Gwo-Bin Lee *acd
aDepartment of Power Mechanical Engineering, National Tsing Hua University, Hsinchu, 30013 Taiwan. E-mail: gwobin@pme.nthu.edu.tw; Tel: +886 3 5715131 Ext. 33765
bDepartment of Electrical Engineering, National Tsing Hua University, Hsinchu, 30013 Taiwan. E-mail: hp@ee.nthu.edu.tw; Tel: +886 3 5162206
cInstitute of NanoEngineering and Microsystems, National Tsing Hua University, Hsinchu, 30013 Taiwan
dInstitute of Biomedical Engineering, National Tsing Hua University, Hsinchu, 30013 Taiwan

Received 12th August 2019 , Accepted 9th October 2019

First published on 10th October 2019

Emerging and re-emerging infectious diseases pose global threats to human health. Although several conventional diagnostic methods have been widely adopted in the clinic, the long turn-around times of “gold standard” culture-based techniques, as well as the limited sensitivity of lateral-flow strip assays, thwart medical progress. In this study, a smartphone-controlled, automated, and portable system was developed for rapid molecular diagnosis of pathogens (including viruses and bacteria) via the use of a colorimetric loop-mediated isothermal amplification (LAMP) approach on a passive, self-driven microfluidic device. The system was capable of 1) purifying viral or bacterial samples with specific affinity reagents that had been pre-conjugated to magnetic beads, 2) lysing pathogens at low temperatures, 3) executing isothermal nucleic acid amplification, and 4) quantifying the results of colorimetric assays for detection of pathogens with an integrated color sensor. The entire, 40 min analytical process was automatically performed with a novel punching-press mechanism that could be controlled and monitored by a smartphone. As a proof of concept, the influenza A (H1N1) virus and methicillin-resistant Staphylococcus aureus bacteria were used to characterize and optimize the device, and the limits of detection were experimentally found to be 3.2 × 10−3 hemagglutinating units (HAU) per reaction and 30 colony-forming units (CFU) per reaction, respectively; both such values represent high enough sensitivity for clinical adoption. Moreover, the colorimetric assay could be both qualitative and quantitative for detection of pathogens. This is the first instance of an easy-to-use, automated, and portable system for accurate and sensitive molecular diagnosis of either viruses or bacteria, and it is envisioned that this smartphone-controlled apparatus may serve as a platform for clinical, point-of-care pathogen detection, particularly in resource-limited settings.


Pathogens are microorganisms or specific agents that cause human diseases and can be classified as bacteria, viruses, fungi, parasites, or prions.1,2 Throughout human history, emerging infectious diseases (e.g., the human immunodeficiency virus (HIV) that began its global spread in the early 1980s) have posed serious threats to world health. For instance, influenza pandemics cause 250[thin space (1/6-em)]000 to 500[thin space (1/6-em)]000 deaths worldwide each year.3 Currently, more than 1400 microorganisms have been found to cause infectious diseases in humans,4,5 and over 6.7 million deaths worldwide occur annually as a result of these pathogens (based on 2018 World Health Organization statistics6). Given this, rapid and accurate diagnosis of infectious disease-causing agents is of great demand for preventing their spread and reducing mortality in patients that have already succumbed to diseases.

Currently, there are four approaches for viral or bacterial detection, with microbial culture being the “gold-standard.” Although it is the most commonly used in hospitals, the analytical time for pathogen identification via culture is at least three days;7 given such time requirements, as well as the labor-intensive procedures and large-scale equipment requirements associated with microbial culture, point-of-care (POC) diagnostics dependent on microbial culture are inherently unrealizable. Alternatively, enzyme-linked immunosorbent assays (ELISA), which rely on antibody–antigen reactions, require only 0.5–2 h to detect pathogens. However, several studies have reported that some polyclonal antibodies may cross-react with off-target serotypes, leading to false-positive results.8,9 Furthermore, ELISA results are also affected by individuals' immuno-responses (e.g. “windows period” issue). Alternatively, lateral-flow strip assays (LFSA), which are also based on antigen–antibody interactions, have been proven to be rapid (15–30 min) and relatively cheap; unfortunately, their lack of sensitivity and proneness to generating false-negative results in individuals infected with low concentrations of bacteria or viruses have limited their widespread adoption in the diagnostic clinic.10–12

Alternatively, polymerase chain reaction (PCR) is also widely used for pathogen detection nowadays since nucleic acid sequences specific to a wide variety of microbes can be exponentially amplified to detectable levels in only two hours or less. Furthermore, the PCR assays are generally highly sensitive as well.13 However, PCR requires bacteria/virus isolation and nucleic acid extraction, as well as bulky and expensive thermocycling equipment. In addition, the 2 h thermocycling period for traditional PCR may be unacceptably long for certain POC applications.14 As such, several isothermal amplification approaches have been developed in recent years; among them, loop-mediate isothermal amplification (LAMP) may be the most promising alternative to PCR since it is the most sensitive and specific of the isothermal amplification approaches.15–17 Briefly, LAMP utilizes four to six primers that recognize distinct sequences on the target nucleic acid molecule, and 109 copies of a target DNA molecule can be generated within 1 hour; it is also relatively tolerant to common reaction inhibitors, meaning that quick and simple nucleic acid extraction approaches can be used.18–20 Furthermore, LAMP does not require initial thermal denaturing due to the high strand-displacement activity of Bacillus stearothermophilus (Bst) DNA polymerase. It can also be used for the detection of RNA provided that a complementary DNA intermediary has first been generated by reverse transcription.21,22 Since LAMP offers such advantages in terms of specificity, sensitivity, reaction efficiency, and product yield, it may be a promising tool for carrying out rapid and accurate pathogen diagnosis.

Recently, microfluidic devices have advanced the detection of pathogens when incorporated with LAMP.23–26 Among them, capillary-driven, LAMP-based LFSAs have been demonstrated as promising tools for POC diagnosis due to the fact that they offer the benefits of portability, low cost, high speed, and low power consumption. However, these methods still have limitations, and most still require large-scale equipment for subsequent analysis. For instance, a multi-layer microfluidic device equipped with filter papers that enabled sample pre-treatment and LAMP amplification for Escherichia coli (E. coli) detection was developed.27 Although the limit of detection (LOD) was as low as 5 bacteria/reaction within 70 min, an oven and a hand-held ultraviolet (UV) emission source were required for amplification and detection. Similarly, a LAMP-based, paper-based colorimetric LFSA utilizing gold nanoparticles for detection of dengue viruses was reported.28 The LOD of this assay was found to be 103 copies/reaction, and a portable, battery-powered heating device was used for amplification and detection. However, this approach still required off-chip sample pre-treatment, which may not be acceptable for POC applications.

An integrated microfluidic system combined with the LAMP technique was also reported for detection of methicillin-resistant Staphylococcus aureus (MRSA).26 The device was capable of automatically carrying out all diagnostic processes (bacteria isolation via magnetic beads, DNA extraction via thermal lysis, LAMP amplification, and optical detection) within 65 min, and the LOD was as low as 1 CFU per reaction. However, liquid transport in the chip and optical detection were achieved by large and expensive components such as vacuum pumps, laser modules, and photo-multiplier-tubes; additionally, a laptop computer was required for automatic control, potentially hindering its adoption in resource-limited settings. To circumvent these issues, a smartphone-based, “sample-to-answer,” portable system integrated with a self-driven microfluidic device was developed herein for rapid detection of two common, potentially life-threatening pathogens, including influenza A (InFA, H1N1) and MRSA. The pathogen sample was loaded into a passive, self-driven microfluidic chip capable of carrying out all analytical steps for on-site pathogen detection. With an aid of a punching-press mechanism from the custom-made portable device, the liquid flow could be regulated orderly and mixed well. Moreover, the entire procedure was automatically, remotely controlled and monitored with a smartphone such that this device could be simply operated by non-specialist personnel. As a proof of concept, the entire analytical assay for detection of influenza A and MRSA was successfully implemented in this portable system. Experimental data showed that it was rapid (<40 min) and more compact in size (19.3 × 16.5 × 12.3 cm) and the power was supplied by a lithium (Li)-ion battery, making the whole system more suitable for POC applications when compared with previous works.26,27 Furthermore, the colorimetric results of the LAMP assay could be quantified by an integrated color sensor and displayed on the smartphone APP such that the human-error issue by naked eyes could be eliminated. This is the first time that an automated portable system capable of rapid pathogen detection, including pathogen isolation, DNA extraction, colorimetric LAMP reaction, and quantitative optical detection, has been demonstrated on a compact system, which is amenable for clinical use in the resource-limited setting in the near future.


Overview of the approach

The entire procedure for detection of pathogens with the portable system developed herein has been schematically illustrated (Fig. 1). All reagents were prepared off-chip and loaded into the respective reservoirs of the chip prior to detection. The liquids spontaneously filled the microchannels via capillary force and were stopped by hydrophobic soft valves, which could be triggered on demand with a punching-press mechanism equipped on the device. The color intensity (a proxy for viral or bacterial load) was measured by a color sensor, and the entire analytical process was automatically performed within 40 min under the control of a smartphone. Each step mentioned above is outlined in detail below.
image file: c9lc00797k-f1.tif
Fig. 1 Experimental workflow of a sample-to-answer, potable platform for rapid pathogen detection.

Virus and bacteria strains

Seven pathogens were used in this study, including 1–2) two inactivated influenza A virus strains (H1N1: InfA/H1 [97N510H1] and H3N2: InfA/H3 [A/California/7/2004]), 3) an influenza B (InfB) virus strain (InfB: 94N399IB) provided by the Department of Microbiology and Immunology of National Cheng Kung University (Taiwan), 4) a MRSA strain from the Bioresource Collection and Research Center (BCRC; BCRC 15211), 5) a Staphylococcus aureus (SA) strain from the American Type Culture Collection (ATCC, ATCC 25923), 6) an E. coli strain (ATCC 25922), and 7) a clinical Acinetobacter baumannii strain from Chang Gung Hospital (Taiwan; IRB No. 104-0117B). All inactivated viruses obtained from the above organization were stored at −80 °C prior to use. The virus concentrations were measured to be 32 HAU by a hemagglutination assay,29 and the concentrations of the bacteria were estimated by colony counts on agar plates via a serial dilution technique.30 Note that all bacteria strains were freshly cultured in Luria-Bertani broth (L3152, Sigma, USA) under overnight rotation at 200 rpm at 37 °C in an incubator (LTI-601, TKS, Taiwan) to standardize the initial concentration and then serially diluted 10 folds (101–105 fold) to determine the LOD without a separate culture step. For optimization of the colorimetric assay, viral RNA and bacterial DNA were extracted directly by using a thermal-lysis method developed previously.31,32 The concentration of viral RNA was determined with a NanoDrop ultraviolet (UV) spectrophotometer (DU530 UV/Vis, Beckman, USA), and viral DNA copy numbers were estimated based on a protocol reported in a previous work.33

Preparation of affinity molecule-conjugated magnetic beads

In order to isolate influenza A viruses and MRSA bacteria from specimens, a single-stranded H1N1-specific aptamer screened in a previous study34 (sequence was listed in Supplementary Table S1) and the antibiotic vancomycin (V2002-100MG, Sigma), respectively, were conjugated onto the surfaces of magnetic beads. For the former, the 5′ end of the aptamer was modified with an amine group and then covalently bonded with magnetic beads (stock concentration = 107 beads per μL, 1.05 μm diameter, Dynabeads® MyOne™ carboxylic acid, Invitrogen, USA) by carboxylation.35 For the latter, vancomycin was conjugated on the surface of nano-magnetic beads (stock concentration = 1.4 × 108 beads per μL, AllMag™ PM3-020, 180 nm diameter, So-Fe Biomedicine, China) via a polyethylene glycol-diamine (753084-1G, Sigma, USA) linkage method.31 Both affinity molecule-conjugated beads were stored at 4 °C prior to use.

Colorimetric LAMP assay

Visual detection based on a color change under ambient light is the most promising detection means for POC applications given its simplicity and the lack of need for sensitive, expensive instruments. Hydroxynaphthol blue (HNB) is one of several metal ion indicators that is regularly used as a visual dye for the LAMP technique,36,37 and it is characterized by a violet to sky blue transition in successful LAMP reactions. However, this shift only occurs under concentrations of Mg2+ (>6 mM).36,37 To limit the need for such high concentrations herein, we proposed to add Mn2+ and calcein38 in reaction mixtures. A LAMP reaction mixture (25 μL) was prepared off-chip prior to experiments after designing primers that would target fragments of hemagglutinin (HA) and a methicillin-resistance (mecA) gene for identification of the H1N1 virus and MRSA, respectively. Sequences for these genes were obtained from the National Center for Biotechnology Information (NCBI) website, and primers (ESI Table S1) were designed with PrimerExplorer V5 software (Eiken Chemicals Corporation, Tokyo, Japan). To shorten the reaction times, three additional sets of loop primers (LF and LB) were designed for each assay, and a real-time LAMP fluorescent detection approach was adopted to determine the best set of primers on a real-time PCR system (StepOne™, Applied Biosystems, USA) using the following components in the reaction mixtures: 1.25 μL of deionized distilled water (ddH2O), 2.5 μL of 10× isothermal amplification buffer II (New England Biolabs, USA), 4 μL of 5 M betaine (Sigma, USA), 1.5 μL of magnesium sulfate (2–8 mM; New England Biolabs., USA), 3.5 μL of 10 mM deoxynucleotide (dNTP) solution (Protech, Taiwan), 0.5 μL each of 10 μM F3 and B3 primers (outer primers), 1.5 μL each of 20 μM forward internal primers (FIP) and backward internal primers (BIP) primers, 1 μL each of 10 μM loop forward primers (LF) and loop backward primers (LB), 3 μL of 2 mM HNB (Sigma, USA), 1.25 μL of 20× concentrated calcein solution (containing 0.5 mM calcein and 5 mM manganese chloride; Sigma, USA), 1 μL of Bst DNA polymerase (M0374, 8 U μL−1, New England Biolabs, USA), and 1 μL of pathogen samples. It is worth noting that betaine was experimentally found to inhibit the nucleic acid amplification of the mecA gene (ESI Fig. S1), and therefore water was added in place of betaine for this LAMP assay.

Magnesium (Mg2+) concentrations of 2, 3, 4, 5, 6, 7, and 8 mM (as magnesium sulfate) were tested in triplicate with the aforementioned reaction components, incubated at 62 °C (MRSA) or 63 °C (H1N1) for 20–50 min and followed by 80 °C for 5 min to terminate the reaction; the fluorescent signal was recorded every 30 s and analyzed by StepOne software v2.3 (Applied Biosystems, USA). Finally, the colorimetric results from the real-time LAMP assays were recorded with a digital camera (D40X, Nikon, Japan).

Chip design and fabrication

The passive, self-driven microfluidic chip was consisted of a polydimethylsiloxane (PDMS) microfluidic structure layer, a hydrophilic film, a hydrophobic PDMS layer, and a glass substrate (Fig. 2a). The PDMS microfluidic structure layer was designed with AutoCAD and SolidWorks software and then fabricated utilizing a computer-numerical-control (CNC) machining process (EGX-400, Roland Inc., Japan) with a 0.5 mm-diameter drill actuated at rotation rates and feed speeds of 27[thin space (1/6-em)]000 rpm and 6 mm s−1, respectively;39 afterwards, a PDMS soft lithography technique was applied to replicate the chip with the inverse patterns.40 High drill rotational rates and low feed speeds tended to result in smoother master molds, such that the replicated PDMS structure and PDMS hydrophobic layers could be bonded with ease.
image file: c9lc00797k-f2.tif
Fig. 2 (a) Exploded view of the self-driven microfluidic chip consisting of a PDMS microfluidic structure layer, a hydrophilic film, a PDMS hydrophobic layer, and a glass substrate. (b) A photograph of the chip. (c) Schematic illustration of the chip design. The area enclosed by the red dotted line represents the LAMP reaction module (module B), with the other area being used for sample treatment (module A).

Next, a thin, hydrophobic layer of PDMS (∼50 μm) was spin-coated (700 rpm for 30 s) on a microscope glass slide (75.2 × 25.2 × 1 mm), cured at 80 °C for 2 h, partially covered by a hard mask, and exposed to oxygen plasma treatment (CUTE MP/R, Femto Science, Korea). Afterwards, the partially activated PDMS surface on the glass slide was spin-coated with a uniform layer of an UV glue film (∼11 μm, GL-03, Coretronic Corporation, Taiwan) and subsequently placed into a vacuum box. After purging with nitrogen to remove oxygen gas, the box was vacuumed and exposed to UVA light (365 nm, 13 W) for 30 min, resulting in a hydrophilic-patterned slide (ESI Fig. S2).41,42 Finally, this slide was bonded to the upper PDMS microfluidic structure layer via oxygen plasma treatment, such that a passive microfluidic chip capable of liquid self-transportation could be completely assembled (Fig. 2b).

To carry out precise and rapid detection of viruses and bacteria, the chip was equipped with two modules: a sample pre-treatment module and a LAMP reaction module (Fig. 2c). The former was consisted of four open reservoirs (depth = 2 mm) for storing samples, magnetic beads, washing buffer, and lysis buffer (reservoir diameters of 4, 4, 6, and 5 mm, respectively). The widths and heights of the liquid channels were 600 μm and 500 μm, respectively. In addition, an outlet chamber was responsible for collecting the wastes at the end of the channel. Moreover, a capillary pump and four pieces of filter paper (No. 5C, Advantec, Japan) were placed in the outlet chamber to facilitate transport of liquid, such that the liquid wastes could be collected with ease.

The LAMP reaction module was comprised of individual zones for the negative control, positive control, and test sample (6 mm in diameter and 0.5 mm in height). Note that the standardization of diagnosis is of great importance in precision medicine and therefore both positive and negative controls are critical indications for validation and verification of diagnosis in clinical applications.43 The sample testing zone was connected to the pre-treatment reaction by a serpentine-shaped channel; this channel served as a relatively high fluidic resistor for reducing liquid backflow during the LAMP incubation since it was relatively narrow (400 μm), long, and featured a number of right-angle turns. With this new approach, samples and reagents could be precisely regulated.

Furthermore, hydrophobic soft valves were inserted in the downstream end of each reservoir to block liquid flow; each could be individually activated by a punching-press mechanism. This design permitted temporary incubation of the samples during the pre-treatment stages (e.g., virus/bacteria isolation). The maximum widths of the dual- and single-inlet hydrophobic soft valves were 3400 and 1900 μm, respectively, while the height of all valves was 550 μm. Note that their height was designed to be 250 μm higher than the hydrophilic channel, such that the liquid flow could be stopped effectively.41 With this new design, liquid flow could be stopped and resumed by a simple press, such that the entire analytical process could be performed successfully with a programmed punching press mechanism on a portable device. Note that the capability of spontaneous liquid motion by wetting property of the hydrophilic film has been demonstrated previously.42

Experimental procedures

The experimental procedure was schematically illustrated in Fig. 3. As mentioned above, all reagents were loaded into the corresponding reservoirs on-chip before performing the assay, including 20 μL of the pathogen sample, 20 μL of affinity molecule-conjugated magnetic beads (H1N1: 5 × 105 beads per μL; MRSA: 1.4 × 107 beads per μL), 25 μL of ddH2O, 20 μL of lysis buffer (H1N1: Favnk 001 (0.5×), Favorgen, Taiwan; MRSA: lysostaphin, 10 ng μL−1, Sigma, USA), and 25 μL of LAMP reaction mixtures. The samples and magnetic beads were brought into contact first by mechanically pressing the hydrophobic soft valve; they were then self-transported to the reaction chamber and incubated at room temperature (RT) for 5 min. Simultaneously, a vertical vibrating agitation was applied while pressing the hydrophobic soft valve, which resulted in a highly effective mixing (described below). Following that, an external magnet was employed to collect the magnetic bead-pathogen complexes. After washing out the unbound wastes, pathogens were chemically lyzed at RT (H1N1) or 37 °C (MRSA) for 5 min and then transported to the downstream chamber to be mixed with the LAMP reagents for 15 min (H1N1; 63 °C) or 20 min (MRSA: 62 °C) incubation. The results of the colorimetric assay38 were analyzed by an integrated color sensor and wirelessly displayed on the mobile APP (described below). The entire procedure was automatically performed within 40 min, which was much faster than traditional real-time PCR approaches.44,45 Note that the reagents could be pre-packaged into the chip via a polyethylene/thermoplastic elastomer film,46 such that the manual sample and reagent dispension process could be simplified prior to the detection.
image file: c9lc00797k-f3.tif
Fig. 3 Schematic diagram of the pathogen diagnostic process. (a) Loading of samples, magnetic beads, wash buffer, lysis buffer, and LAMP reagents. (b) Pathogen capture by affinity molecule-conjugated magnetic beads. (c) Collection of pathogen-magnetic bead complexes with an external magnet. (d) Washing out of unbound wastes. (e) Low-temperature lysis. (f) Amplification of specific target sequences by the LAMP reaction. (g) Colorimetric reaction.

Custom-made portable pathogen diagnostic system

The portable system was composed of three devices: a microfluidic chip, a portable control device, and a smartphone. The 19.3 × 16.5 × 12.3 cm, cube-shaped portable control device (Fig. 4) was constructed by a three-dimensional printing technique and featured a neodymium magnet (2000 Gauss) for magnetic bead collection, a Bluetooth transceiver (MD08R-C2A, Hotlife, Taiwan) for wireless connection with the smartphone, two 5-V stepping motors (PM25S-024-ZH89, NMB Inc., USA), two photo-interrupter modules (KY-010, Keyes, China) for chip positioning, a thermal control module (L298, STMicroelectronics, Switzerland) equipped with a Peltier device (TEC-1-4905, TANDE Corp., Taiwan) and a temperature sensor (MAX6675, MAXIM, USA) for temperature regulation, a punching-press mechanism for flow control and mixing, and a color sensor (TCS3414CS, Intel Inc., USA) for detection of colorimetric results. Note that the color sensor was adopted since it provided more accurate results than reading directly with the smartphone camera.47 All sensors and actuators were incorporated with the microcontroller unit (MCU; Arduino Nano, Italy) in order to automate motor control, temperature, detection, and wireless connectivity.
image file: c9lc00797k-f4.tif
Fig. 4 Illustration of the portable system composed of two stepping motors, two sets of gears, two photo-interrupters, a punching-press mechanism, an Arduino-based microcontroller circuit, a thermal control module, a color sensor, and a microfluidic chip.

Since the smartphone has become an essential device for most people, it is seeing growing use in human health monitoring48 and diagnosis.49 Therefore, a mobile APP with a graphical user-interface was designed to operate the system designed herein (ESI Fig. S3). The user interface permits users to 1) adjust reaction times and temperatures (if necessary) and 2) monitor the LAMP reaction status in real-time. After pressing the “START” button in the APP, the request was then sent wirelessly to the MCU. The portable system could therefore execute the programmed functions automatically and remotely.

Mixing efficiency

The mixing efficiency was estimated by calculating the mixing index (σ) as in a prior work.39 Briefly, 20 μL each of blue ink and water were loaded into the reservoirs. While the valve was actuated by the punching-press mechanism, a vertically oriented vibration was applied simultaneously by repeating the punching procedure, thus mixing the liquids. The images were recorded by a digital camcorder (GZ-HM650BU, JVC, Japan), analyzed with ImageJ software (National Institutes of Health, USA), and the mixing index (σ) was calculated as in prior works.39,50

Results and discussion

Characterization of the custom-made portable system

The performance of the custom-made portable system was evaluated prior to the pathogen detection experiments. Since the LAMP assay requires a constant temperature of 62–63 °C, the temperature profiles of the thermal control module (i.e., LAMP reaction module) were first measured and recorded by a digital thermometer (SD-947, Reed Instruments, USA). The heating rate of the module was experimentally found to be 5.2 °C s−1 and the temperature could be well-regulated with a low temperature fluctuation (within ±1.5 °C) over a period of 40 min (ESI Fig. S4); this level of precision is acceptable for LAMP reactions and is superior to other low-cost, portable heating devices (e.g., pocket warmer51 or instant heat pack52) that have been used for LAMP reactions (i.e. heating rate = 2 °C min−1). Furthermore, insulation (e.g., Styrofoam boxes) was required to generate consistent heating in these other devices. Given this comparison, as well as the high flow precision actuated by the punching press mechanism's fine-tuned triggering of the hydrophobic soft values (Fig. 5a–c), the system designed herein appears to be more amenable to POC diagnostics.
image file: c9lc00797k-f5.tif
Fig. 5 Actuation of hydrophobic soft valves utilizing the punching-press mechanism. (a) Initiation, (b) actuation, and (c) valve opening. (d) The mixing principle during the punching-press process. (e) The images of the mixing profile with/without mixing.

Generally, passive micromixers were equipped on capillary-driven microfluidic devices to achieve mixing of liquids via diffusion through microchannels with particular geometric designs.53 Although passive micromixers are easy to integrate into chips with advantages of simple operation and no need for bulky equipment, they only achieve high mixing efficiencies across relatively long channels (i.e., when there is a long contact area and mixing time).54 In contrast, active micromixers offer rapid and effective mixing;39 however, they require an external power supply,55 which may impede their miniaturization for low-cost, POC diagnostics.

Given this, a new mixing approach utilizing an automatic, mechanical punching-press mechanism (with repeated presses made at a frequency of 6.9 Hz) was demonstrated in this work (Fig. 5d), and this system resulted in mixing indices that increased from 28.9% to 98.6% within 1 s (Fig. 5e), which was comparable to the ones from the other active micromixers (e.g. pneumatically-driven micromixers32,39,50 and acoustically-driven micromixers56,57). With this approach, the fluids could be mixed efficiently as they were transported passively downstream. Thus, this unique design may be more favorable for POC applications, particularly in resource-limited settings.

Optimization of the colorimetric LAMP assay

By reducing colorimetric LAMP reaction and measurement times, their clinical utility could be greatly improved. Herein optimization of the colorimetric LAMP assay comprised three steps, including 1) screening multiple sets of loop primers, 2) selecting the optimal concentration of Mg2+ within the LAMP mixture, and 3) determining optimal visual detection conditions with the optimal conditions established from steps 1 and 2. Three sets of loop primers for detection of the H1N1 virus and MRSA were evaluated by utilizing real-time fluorescence detection in the LAMP assay (ESI Table S1), and loop primers infA-LF and infA-LB-2 were characterized by the earliest detection signals when used with the influenza A virus, and loop primers mecA-LF-1 and mecA-LB were the optimal primers for the detection of MRSA (Fig. 6a and b). Therefore, these two sets of loop primers were used for all subsequent experiments.
image file: c9lc00797k-f6.tif
Fig. 6 Optimization of the LAMP assay. (a and b) The effect of three sets of loop primers on detection of (a) H1N1 and (b) MRSA. (c and d) Real-time LAMP reaction with different concentrations of Mg2+ for detection of (c) H1N1 viruses and (d) MRSA. LF: Loop forward primers, LB: loop backward primers, and NC: negative control.

Furthermore, seven concentrations of Mg2+ were further tested (Fig. 6c and d). From the results of real-time fluorescence detection, the LAMP products were detected much earlier with the addition of Mg2+ concentrations of 5 and 6 mM for the detection of MRSA and H1N1 virus, respectively. Therefore, the optimal concentration of Mg2+ was determined to be 5 mM for MRSA and 6 mM for H1N1 virus for the subsequent visual detection.

Once LAMP primers and concentration of Mg2+ had been optimized, the visual properties were further explored. As shown in Fig. 7a, the positive controls (H1N1: ∼103 copies; MRSA: 3 × 103 CFU) turned to blue after 15 (H1N1) or 20 min (MRSA) of LAMP reaction. It is worth noting that with an extended reaction time (H1N1: 30 min; MRSA: 40 min), the color intensity dissipated appreciably due to the formation of HNB–Mn2+ complexes.38 As polymerization proceeds in the LAMP reaction, the color of HNB changes according to the depletion of free Mn2+ ions. Therefore, the negative control (without generating amplicons) showed a complete color change from purple to orange, as expected,38 which was readily distinguishable from the positive control (Fig. 7a). To the best of our knowledge, this is the first time that Mn2+ has been found to play dual roles as a replacement to Mg2+ to chelate the negative charges of the dNTPs36 (i.e. enhance the color intensity) and formation of HNB–Mn2+ complexes (deplete the color intensity). Given this, the optimal visual detection dye comprised Mn2+ ions (final concentrations for H1N1 and MRSA of 6 and 5 mM, respectively), calcein (final concentration: 25 μM), and HNB (final concentration: 0.5 mM), and these conditions were used for all subsequent reactions.

image file: c9lc00797k-f7.tif
Fig. 7 Images of colorimetric LAMP assays. (a) Comparison of color changes in LAMP assays with different dyes for 20 and 40 min. I: HNB dye with 5 mM of Mg2+ ion, II: mixed dye with 5 mM of Mg2+ ion for the detection of MRSA, and III: mixed dye with 6 mM of Mg2+ ion for detection of the H1N1 viruses. N: Negative control and P: positive control. (b) Images of the sensitivity test utilizing optimized colorimetric LAMP reaction mixtures for detection of the H1N1 viruses within 15 and 30 min. (c) Images of the sensitivity test utilizing optimized colorimetric LAMP reaction mixtures for the detection of MRSA bacteria within 20 and 40 min. (d) The relationship between the optical signals and the concentrations of H1N1 viral RNA from the color sensor. (e) The relationship between the optical signals and the concentrations of MRSA from the color sensor. The dotted line represents the threshold, which was calculated by subtracting two standard deviations from the mean value of the negative control signal. Error bars represent standard deviation (n = 3).

Sensitivity and specificity of the colorimetric LAMP assay

The sensitivity of the colorimetric LAMP assay was then investigated by using 10-fold serial dilutions of the H1N1 virus (stock concentration = 32 HAU) and MRSA bacteria (stock concentration = 3 × 105 CFU μL−1), and positive signals (blue) were observed when concentrations were higher than 3.2 × 10−3 HAU per reaction for H1N1 virus and 30 CFU per reaction for MRSA, respectively, within 15 and 20 min, respectively (Fig. 7b and c). Note that the specificity of the colorimetric assay has been verified utilizing seven different pathogens with the incorporation of specific target primer sets. Since only target genes (HA and mecA) could be recognized and successfully amplified with clear LAMP products, the positive signals (blue color) was only achieved by the H1N1 virus and MRSA, while the colors for other pathogens were the same as the negative control (ESI Fig. S5a and b). Agarose gel electrophoresis also indicated that the ladder-like LAMP products were only observed in samples containing the H1N1 virus or MRSA, thereby demonstrating the assay's high specificity (ESI Fig. S5c and d).

Moreover, the color intensity dissipated as a function of pathogen concentration and LAMP reaction time. As the concentration of pathogens decreased, the degree of color dissipation increased (Fig. 7b and c). Furthermore, the colorimetric results were quantified by a color sensor, and a negative linear association between the measured optical signals (ΔEab) and the concentrations of the H1N1 virus and MRSA (after 30 or 40 min of reaction time, respectively) was revealed (Fig. 7d and e). According to these results, it is evident that the detection of the H1N1 virus and MRSA bacteria could be carried out within 15 or 20 min, and the LODs were 3.2 × 10−3 HAU per reaction and 30 CFU per reaction, respectively. Moreover, with a longer reaction times, not only did the LOD of the colorimetric LAMP assay decrease by one order of magnitude, but the assay also exhibited a capacity for qualitative and quantitative diagnosis that was comparable to fluorescence detection58,59 and turbidity approaches;60,61 as the latter failed to produce any visible turbidity for samples with low concentrations of pathogens.37 It is worth noting that the concentration of metallic ions in the LAMP reaction mixture should be precisely controlled in each assay to ensure that the quantitative colorimetric assay yields unbiased quantitative results. However, the concentration of these ions may be vary as a result of having used lysis buffers of variable salt concentrations in the DNA/RNA extraction process; this may affect color changes and thus compromise the accuracy of the quantification. This issue could be minimized by using complementary DNA probe-conjugated magnetic beads after lysis in order to collect target DNA molecules (and thereby leaving behind the salts from the extraction buffer). Although this optimized colorimetric LAMP assay have such limitation to achieve its accuracy of the quantification, it could still yield a visibly distinct color change in only 15 min for H1N1 and 20 min for MRSA. Therefore, on-chip LAMP reaction times of 15 and 20 min were set, respectively.

The limit of detection of the custom-made portable system

The LODs for the diagnosis of influenza A and MRSA were investigated herein with the optimized colorimetric LAMP assay conditions on the developed portable system. Pathogens at concentrations ranging from 3.2 × 100 to 3.2 × 10−5 HAU per reaction for H1N1 viruses and 3 × 100 to 3 × 104 CFU per reaction for MRSA bacteria were analyzed on-chip, and the colorimetric results were quantified with the color sensor and displayed on the smartphone APP. A purple-to-blue color change only appeared in 1) the positive controls and 2) samples with concentrations higher than 3.2 × 10−3 HAU per reaction for H1N1 viruses or 30 CFU per reaction for MRSA (Fig. 8a), which is consistent with the optical signals (ΔEab) from the color sensor mentioned above. Moreover, these LODs are comparable to those of prior studies of influenza A62,63 and MRSA.64,65 When compared to the LFSA approaches, this microfluidic system has a higher sensitivity to achieve a lower LOD within 40 min.66,67 In contrast, LFSAs have high LODs66,68 (e.g., 104–106 virus particles for influenza A) or additional enrichment steps67,69 (e.g., 2 CFU with a 48 h incubation for MRSA)67,69 to overcome this limited sensitivity.
image file: c9lc00797k-f8.tif
Fig. 8 The limit of detection on the developed portable system. (a) Images of the colorimetric assay for detection of influenza A (H1N1) viruses. (b) Images of the colorimetric assay for detection of MRSA. (c) The relationship between the output signals and the concentration of the H1N1 virus. (d) The relationship between the output signals and the concentration of MRSA. NC: Negative control, PC: positive control. The dotted line represents the threshold, which was calculated by subtracting two standard deviations from the mean value of the negative control signal. Error bars represent standard deviation (n = 3).


In this study, a portable, sample-to-answer system for rapid, accurate, and sensitive molecular detection of either viruses or bacteria was developed. This novel system was capable of 1) performing a colorimetric molecular diagnosis on a passive, self-driven microfluidic chip, 2) executing the procedure with a programmed punching-press mechanism, 3) detecting the results with a color sensor, and 4) controlling and monitoring the whole process with a smartphone. The entire process including pathogen purification, nucleic acid extraction, colorimetric isothermal nucleic acid amplification, and optical detection could be automatically performed within 35–40 min, and the LODs were only 3.2 × 10−3 HAU per reaction for H1N1 viruses and 30 CFU per reaction for MRSA. This platform could be applied for the detection of other viruses and bacteria as well, if suitable LAMP primers are used. Briefly, the prototype was affordable, sensitive, specific, user-friendly, rapid, robust, deliverable to end users, and no need for complex equipment, which followed the POC criteria established by WHO. However, there are still some issues that need to be addressed for development of a successful “real-world” product. For instance, multiple reagent dispension processes with human invention may hinder the POC setting. Thus, all necessary reagents should be integrated into a “sample-to-answer” cartridge with a constant and reproducible manner (e.g. lyophilisate) to ensure that the assay requires only the addition of sample in the future. Additionally, the clinical specimen will be further tested to validate this developed system in the near future. Moreover, the developed system should be benchmarked with commercial established systems before marketing. After the developed system may become a useful tool for POC applications, particularly in the resource-limited settings of developing countries.

Author contribution

Y. D. Ma designed the microfluidic chips, implemented the experiments, and prepared the manuscript. K. H. Li designed the mobile APP and programmed the MCU codes. Y. H. Chen was involved in construction of the automatic system and trouble-shooting. Y. M. Lee designed, constructed, and evaluated the electrical control circuit. S. T. Chou designed the portable device. Y. Y. Lai was involved in development of color intensity algorithm. P. C. Huang supervised the electrical control circuit design, provided valuable suggestions for trouble-shooting and proofread the manuscript. H. P. Ma supervised the works of electrical control circuit and mobile APP interface design and proofread the manuscript, G. B. Lee proposed the concept of the entire system, supervised the experiments, and proofread the manuscript.

Conflicts of interest

There are no conflicts to declare.


The authors would like to acknowledge financial support from the Ministry of Science and Technology (MOST) of Taiwan (MOST 107-2221-E-007-013-MY3 and 106-2221-E-007-029-MY3). Partial financial support from the “Higher Education Sprout Project” of Taiwan's Ministry of Education (Grant No.108Q2713E1) is also greatly appreciated. The authors also thank Dr. Chih-Peng Chang for providing the H1N1 virus. Finally, the authors thank Drs. Huey-Ling You and Mel S. Lee for providing E. coli and MRSA.


  1. B. Alberts, A. Johnson, J. Lewis, M. Raff, K. Roberts and P. Walter, Molecular biology of the cell, Garland Science, New York, 4th edn, 2002 Search PubMed.
  2. S. B. Prusiner and D. T. Kingsbury, CRC Crit. Rev. Clin. Neurobiol., 1985, 1, 181–200 CAS.
  3. M. Richard, M. de Graaf and S. Herfst, Future Virol., 2014, 9, 513–524 CrossRef CAS PubMed.
  4. S. I. Hay, K. E. Battle, D. M. Pigott, D. L. Smith, C. L. Moyes, S. Bhatt, J. S. Brownstein, N. Collier, M. F. Myers, D. B. George and P. W. Gething, Philos. Trans. R. Soc. London, Ser. B, 2013, 368, 20120250 CrossRef PubMed.
  5. C. I. Paules, R. W. Eisinger, H. D. Marston and A. S. Fauci, Ann. Intern. Med., 2017, 167, 805–811 CrossRef PubMed.
  6. World Health Organization, Global status report on alcohol and health 2018, 2018 Search PubMed.
  7. J. Song, C. Liu, M. G. Mauk, S. C. Rankin, J. B. Lok, R. M. Greenberg and H. H. Bau, Clin. Chem., 2017, 63, 714–722 CrossRef CAS PubMed.
  8. A. Macneil, Z. Reed and P. E. Rollin, PLoS Neglected Trop. Dis., 2011, 5, e1175 CrossRef CAS PubMed.
  9. F. Ackermans, A. Pini, D. Wachsmann, M. Scholler, J. Ogier and J. P. Klein, Clin. Exp. Immunol., 1991, 85, 265–269 CrossRef CAS PubMed.
  10. T. M. Uyeki, Pediatr. Infect. Dis. J., 2003, 22, 164–177 Search PubMed.
  11. C. Chartrand, M. M. Leeflang, J. Minion, T. Brewer and M. Pai, Ann. Intern. Med., 2012, 156, 500–511 CrossRef PubMed.
  12. C. Pohlmann, I. Dieser and M. Sprinzl, Analyst, 2014, 139, 1063–1071 RSC.
  13. K. B. Mullis and F. A. Faloona, Methods Enzymol., 1987, 155, 335–350 CAS.
  14. M. Parida, J. Shukla, S. Sharma, S. Ranghia Santhosh, V. Ravi, R. Mani, M. Thomas, S. Khare, A. Rai, R. Kant Ratho, S. Pujari, B. Mishra, P. V. Lakshmana Rao and R. Vijayaraghavan, J. Mol. Diagn., 2011, 13, 100–107 CrossRef CAS PubMed.
  15. Q. Zhu, Y. Gao, B. Yu, H. Ren, L. Qiu, S. Han, W. Jin, Q. Jin and Y. Mu, Lab Chip, 2012, 12, 4755–4763 RSC.
  16. T. Notomi, H. Okayama, H. Masubuchi, T. Yonekawa, K. Watanabe, N. Amino and T. Hase, Nucleic Acids Res., 2000, 28, E63 CrossRef CAS PubMed.
  17. G. T. Walker, M. S. Fraiser, J. L. Schram, M. C. Little, J. G. Nadeau and D. P. Malinowski, Nucleic Acids Res., 1992, 20, 1691–1696 CrossRef CAS PubMed.
  18. Y. P. Wong, S. Othman, Y. L. Lau, S. Radu and H. Y. Chee, J. Appl. Microbiol., 2018, 124, 626–643 CrossRef CAS PubMed.
  19. L. D. Thiessen, T. M. Neill and W. F. Mahaffee, PeerJ, 2018, 6, e4639 CrossRef PubMed.
  20. A. Das, S. Babiuk and M. T. McIntosh, J. Clin. Microbiol., 2012, 50, 1613–1620 CrossRef CAS PubMed.
  21. A. Erice, D. Brambilla, J. Bremer, J. B. Jackson, R. Kokka, B. Yen-Lieberman and R. W. Coombs, J. Clin. Microbiol., 2000, 38, 2837–2845 CAS.
  22. Y. Chander, J. Koelbl, J. Puckett, M. J. Moser, A. J. Klingele, M. R. Liles, A. Carrias, D. A. Mead and T. W. Schoenfeld, Front. Microbiol., 2014, 5, 395 Search PubMed.
  23. Y. Shang, J. Sun, Y. Ye, J. Zhang, Y. Zhang and X. Sun, Crit. Rev. Food Sci. Nutr., 2018, 1–24,  DOI:10.1080/10408398.2018.1518897.
  24. H. Zhang, Y. Xu, Z. Fohlerova, H. Chang, C. Iliescu and P. Neuzil, TrAC, Trends Anal. Chem., 2019, 113, 44–53 CrossRef CAS.
  25. M. Safavieh, M. K. Kanakasabapathy, F. Tarlan, M. U. Ahmed, M. Zourob, W. Asghar and H. Shafiee, ACS Biomater. Sci. Eng., 2016, 2, 278–294 CrossRef CAS PubMed.
  26. S. L. Chen, W. H. Chang, C. H. Wang, H. L. You, J. J. Wu, T. H. Liu, M. S. Lee and G. B. Lee, Microfluid. Nanofluid., 2017, 21, 87–99 CrossRef.
  27. J. T. Connelly, J. P. Rolland and G. M. Whitesides, Anal. Chem., 2015, 87, 7595–7601 CrossRef CAS PubMed.
  28. J. R. Choi, J. Hu, Y. Gong, S. Feng, W. A. Wan Abas, B. Pingguan-Murphy and F. Xu, Analyst, 2016, 141, 2930–2939 RSC.
  29. G. K. Hirst, J. Exp. Med., 1942, 75, 49–64 CrossRef CAS PubMed.
  30. A. Ben-David and C. E. Davidson, J. Microbiol. Methods, 2014, 107, 214–221 CrossRef PubMed.
  31. T. H. Liu, S. S. Cheng, H. L. You, M. S. Lee and G. B. Lee, Analyst, 2019, 144, 1210–1222 RSC.
  32. K. M. Shen, N. M. Sabbavarapu, C. Y. Fu, J. T. Jan, J. R. Wang, S. C. Hung and G. B. Lee, Lab Chip, 2019, 19, 1277–1286 RSC.
  33. C.-H. Tai, Y.-C. Tsai, C.-H. Wang, T.-S. Ho, C.-P. Chang and G.-B. Lee, Microfluid. Nanofluid., 2013, 16, 501–512 CrossRef.
  34. H. C. Lai, C. H. Wang, T. M. Liou and G. B. Lee, Lab Chip, 2014, 14, 2002–2013 RSC.
  35. C. H. Wang, K. Y. Lien, J. J. Wu and G. B. Lee, Lab Chip, 2011, 11, 1521–1531 RSC.
  36. M. Goto, E. Honda, A. Ogura, A. Nomoto and K. Hanaki, BioTechniques, 2009, 46, 167–172 CrossRef CAS PubMed.
  37. A. Das, S. Babiuk and M. T. McIntosh, J. Clin. Microbiol., 2012, 50, 1613–1620 CrossRef CAS PubMed.
  38. B. Pang, S. Yao, K. Xu, J. Wang, X. Song, Y. Mu, C. Zhao and J. Li, Anal. Biochem., 2019, 574, 1–6 CrossRef CAS PubMed.
  39. C. H. Weng, K. Y. Lien, S. Y. Yang and G. B. Lee, Microfluid. Nanofluid., 2010, 10, 301–310 CrossRef.
  40. M. A. Unger, H. P. Chou, T. Thorsen, A. Scherer and S. R. Quake, Science, 2000, 288, 113–116 CrossRef CAS PubMed.
  41. Y. D. Ma, Y. S. Chen and G. B. Lee, Sens. Actuators, B, 2019, 296, 126647 CrossRef CAS.
  42. Y. D. Ma, W. H. Chang, K. Luo, C. H. Wang, S. Y. Liu, W. H. Yen and G. B. Lee, Biosens. Bioelectron., 2018, 99, 547–554 CrossRef CAS PubMed.
  43. C. J. Mattocks, M. A. Morris, G. Matthijs, E. Swinnen, A. Corveleyn, E. Dequeker, C. R. Muller, V. Pratt, A. Wallace and G. EuroGentest Validation, Eur. J. Hum. Genet., 2010, 18, 1276–1288 Search PubMed.
  44. M. Hombach, G. E. Pfyffer, M. Roos and K. Lucke, J. Clin. Microbiol., 2010, 48, 3882–3887 CrossRef PubMed.
  45. S. B. Selvaraju and R. Selvarangan, J. Clin. Microbiol., 2010, 48, 3870–3875 CrossRef CAS PubMed.
  46. F. C. Huang, Y. F. Chen and G. B. Lee, Electrophoresis, 2007, 28, 1130–1137 CrossRef CAS PubMed.
  47. A. Skandarajah, C. D. Reber, N. A. Switz and D. A. Fletcher, PLoS One, 2014, 9, e96906 CrossRef PubMed.
  48. Y. G. Lee, W. S. Jeong and G. Yoon, Telemed. J. E Health, 2012, 18, 585–590 CrossRef PubMed.
  49. T. Laksanasopin, T. W. Guo, S. Nayak, A. A. Sridhara, S. Xie, O. O. Olowookere, P. Cadinu, F. Meng, N. H. Chee, J. Kim, C. D. Chin, E. Munyazesa, P. Mugwaneza, A. J. Rai, V. Mugisha, A. R. Castro, D. Steinmiller, V. Linder, J. E. Justman, S. Nsanzimana and S. K. Sia, Sci. Transl. Med., 2015, 7, 273re271 CrossRef PubMed.
  50. S. Y. Yang, J. L. Lin and G. B. Lee, J. Micromech. Microeng., 2009, 19, 035020 CrossRef.
  51. B. Hatano, T. Maki, T. Obara, H. Fukumoto, K. Hagisawa, Y. Matsushita, A. Okutani, B. Bazartseren, S. Inoue, T. Sata and H. Katano, Jpn. J. Infect. Dis., 2010, 63, 36–40 CAS.
  52. E. C. Yeh, C. C. Fu, L. Hu, R. Thakur, J. Feng and L. P. Lee, Sci. Adv., 2017, 3, e1501645 CrossRef PubMed.
  53. K. Ward and Z. H. Fan, J. Micromech. Microeng., 2015, 25, 94001 CrossRef PubMed.
  54. C. Y. Lee, C. L. Chang, Y. N. Wang and L. M. Fu, Int. J. Mol. Sci., 2011, 12, 3263–3287 CrossRef CAS PubMed.
  55. J. M. Morrissette, P. S. Mahapatra, A. Ghosh, R. Ganguly and C. M. Megaridis, Sci. Rep., 2017, 7, 1800 CrossRef PubMed.
  56. Z. Yang, S. Matsumoto, H. Goto, M. Matsumoto and R. Maeda, Sens. Actuators, A, 2001, 93, 266–272 CrossRef CAS.
  57. D. Ahmed, X. Mao, B. K. Juluri and T. J. Huang, Microfluid. Nanofluid., 2009, 7, 727–731 CrossRef CAS.
  58. T. Kubo, M. Agoh, Q. Mai le, K. Fukushima, H. Nishimura, A. Yamaguchi, M. Hirano, A. Yoshikawa, F. Hasebe, S. Kohno and K. Morita, J. Clin. Microbiol., 2010, 48, 728–735 CrossRef CAS PubMed.
  59. J. Su, X. Liu, H. Cui, Y. Li, D. Chen, Y. Li and G. Yu, BMC Biotechnol., 2014, 14, 8 CrossRef PubMed.
  60. M. Parida, J. Shukla, S. Sharma, S. Ranghia Santhosh, V. Ravi, R. Mani, M. Thomas, S. Khare, A. Rai, R. Kant Ratho, S. Pujari, B. Mishra, P. V. Lakshmana Rao and R. Vijayaraghavan, J. Mol. Diagn., 2011, 13, 100–107 CrossRef CAS PubMed.
  61. Y. Misawa, A. Yoshida, R. Saito, H. Yoshida, K. Okuzumi, N. Ito, M. Okada, K. Moriya and K. Koike, J. Infect. Chemother., 2007, 13, 134–140 CrossRef CAS PubMed.
  62. L. Y. Hung, T. B. Huang, Y. C. Tsai, C. S. Yeh, H. Y. Lei and G. B. Lee, Biomed. Microdevices, 2013, 15, 539–551 CrossRef CAS PubMed.
  63. X. F. Keyi Liu, J. Nanomed. Nanotechnol., 2015, 06, 1 Search PubMed.
  64. Y. H. Liu, C. H. Wang, J. J. Wu and G. B. Lee, Biomicrofluidics, 2012, 6, 34119 CrossRef PubMed.
  65. W. H. Chang, C. H. Wang, C. L. Lin, J. J. Wu, M. S. Lee and G. B. Lee, Biosens. Bioelectron., 2015, 66, 148–154 CrossRef CAS PubMed.
  66. D. A. Green and K. StGeorge, J. Clin. Microbiol., 2018, 56, e00711-18 CrossRef PubMed.
  67. H. Zhang, L. Ma, L. Ma, M. Z. Hua, S. Wang and X. Lu, Int. J. Food Microbiol., 2017, 243, 64–69 CrossRef CAS PubMed.
  68. T. T. Le, P. Chang, D. J. Benton, J. W. McCauley, M. Iqbal and A. E. G. Cass, Anal. Chem., 2017, 89, 6781–6786 CrossRef CAS PubMed.
  69. K. Luo, H. Y. Kim, M. H. Oh and Y. R. Kim, Crit. Rev. Food Sci. Nutr., 2018, 1–14,  DOI:10.1080/10408398.2018.1516623.


Preliminary results within the current study were presented at 1) the 21st International Conference on Miniaturized Systems for Chemistry and Life Sciences (MicroTAS 2017) in Savannah, Georgia, USA (Oct. 22–26, 2017) and 2) the 32nd IEEE International Conference on Micro Electro Mechanical Systems (IEEE MEMS 2019) in Seoul, South Korea (Jan. 27–31, 2019).
Electronic supplementary information (ESI) available. See DOI: 10.1039/c9lc00797k

This journal is © The Royal Society of Chemistry 2019